Pixel-Level Quantification of Subcutaneous Shear-Wave Velocity Heterogeneity for Grading Lymphatic Obstruction in Breast Cancer-Related Lymphedema
Hoe, Z. Y.; Chou, C.-P.; Tzeng, Y.-D.; Ding, R.-S.; Pan, C.-T.; Lee, M.-C.; Lee, C.-H.
Show abstract
Objectives: To develop and validate pixel data analysis (PDA) for reconstructing pixel-level shear-wave velocity (SWV) from saved Holder-Optimized Elastography (HOE) elastograms and to determine whether whole-field subcutaneous heterogeneity provides a quantitative signature of lymphatic obstruction compared with visual high-velocity-area (HVA) counting and sparse manual region-of-interest (ROI) sampling. Materials and Methods: In this cross-sectional study, 110 women after breast cancer surgery underwent lymphoscintigraphy and HOE, a pressure-minimized acquisition framework that preserves conventionally captured SWV information while additionally revealing spatial abnormalities attenuated by handheld acquisition. Reconstruction validation used 200 paired ROIs from 49 PACS-retrieved images spanning nearly the full displayed SWV range; reproducibility was assessed across 96 matched anatomical sites from 6 participants undergoing independently reacquired HOE examinations. PDA, visual HVA counting, and manual ROI sampling were evaluated on the same HOE-acquired elastograms in task-specific common complete-case cohorts. Results: PDA reconstructed machine-reported SWV with near-zero error (mean difference, +0.0009 m/s; mean absolute error, 0.0075 m/s; 95% limits of agreement, -0.0202 to +0.0221 m/s). For subcutaneous variance, interobserver/intraobserver ICCs were 0.718/0.799 for PDA and 0.161/0.560 for manual ROI sampling. In the primary no-versus-partial comparison (n = 93), forearm-pooled PDA subcutaneous variance ratio yielded the highest observed AUC (0.859; 95% CI, 0.772 - 0.947), compared with 0.820 for visual HVA counting and 0.769 for manual ROI sampling. PDA's subcutaneous variance ratio remained a leading feature across all three comparisons (AUC, 0.859 - 0.878), including partial-versus-total grading, where all three frameworks showed high discrimination. Conclusions: PDA accurately reconstructs pixel-level SWV from saved HOE elastograms and enables reproducible whole-field quantification of superficial SWE heterogeneity. Subcutaneous variance ratio was the leading quantitative PDA feature across lymphatic-obstruction comparisons and was particularly informative in partial obstruction. The same approach also enables retrospective quantitative reanalysis of compatible archived elastograms.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- MyoVision-US: an Artificial Intelligence-Powered Software for Automated Analysis of Skeletal Muscle Ultrasonography 92%
- Automated and Manual Quantification of Tumour Cellularity in Digital Slides for Tumour Burden Assessment 92%
- Multi-center Clinical Study Using Optical Coherence Tomography for Evaluation of Cervical Lesions In-vivo 91%
Similar papers in this journal
- Improved accuracy of breast volume calculation from 3D surface imaging data using statistical shape models 92%
- Classification performance bias between training and test sets in a limited mammography dataset 92%
- Silicone Induced Granuloma of Breast Implant Capsule (SIGBIC) diagnosis: Breast Magnetic Resonance (BMR) ability to detect silicone bleeding. 92%
Similar papers in this journal
Similar papers in this journal
- A Machine Learning Ensemble Based on Radiomics to Predict BI-RADS Category and Reduce the Biopsy Rate of Ultrasound-Detected Suspicious Breast Masses 94%
- The NILS study protocol - a retrospective validation study of a preoperative decision-making tool for non-invasive lymph node staging in women with primary breast cancer [ISRCTN14341750] 91%
- Volumetric lung cancer screening reduces unnecessary low-dose computed tomography scans: results from a single-centre prospective trial on 4,119 subjects 89%
Similar papers in this journal
- Artificial Intelligence System Reduces False-Positive Findings in the Interpretation of Breast Ultrasound Exams 93%
- Deep Learning Allows Assessment of Risk of Metastatic Relapse from Invasive Breast Cancer Histological Slides 89%
- An Ensembled Deep Learning Model Outperforms Human Experts in Diagnosing Biliary Atresia from Sonographic Gallbladder Images 89%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.